Deneux, Thomas (2006) Hemodynamic Models: Investigation and Application to Brain Imaging Analysis. PhD thesis ENS/INRIA/ENPC équipe Odyssée, INRIA / ENS / ENPC - équipe Odyssée, EP/X p.201.

Alternative Locations: http://www.imprimerie.polytechnique.fr/Theses/Files/Deneux.pdf

Abstract

An accurate analysis of functional MRI measurements requires a precise understanding of the physiological processes involved in this measure. This PhD work shows both investigations in hemodynamic models and algorithms to use such models in the analysis of brain imaging measurements.
A particular concern with functional MRI is the temporal modeling of the responses to neural activity. Today, the most standard analysis methods use the General Linear Model framework, which supposes a linear relationship between brain activity and the BOLD response. We show how it is possible alternatively to use nonlinear models in data analysis. Our estimation of parameters by energy minimization is the equivalent to linear regression, and our adaptation of the Fisher statistical test enables activation detection, hypothesis testing, and eventually comparison between different models.
We then have extended our methods to the analysis of multiple modalities data, and in particular, proposed a method to estimate the cortical activity underlying simultaneous fMRI and EEG measurements. We were able to achieve accurate estimation on synthetic data.
Additionally to these methodological researches, we have investigated the model equations with an Optical Imaging experiment. We have focused on the dynamic of the blood flow, which is at a crossroad between electrical, metabolism and oxygenation processes. We have identified specific questioning facts about this dynamic, such as nonlinearity with respect to electrical synaptic activity, and delays with respect to the blood volume response. Furthermore, we have conceived a new method for estimating fast erythrocyte motions in the blood from intrinsic optical imaging signal, which might provide a new useful measurement of this blood flow.
In the following synopsis, we summarize the main features of this PhD work, through highlighting the purposes, methods, conclusions and implications of each chapter. Also, we tried to present an objective criticism of this work, through mentioning both its original contributions and its weaknesses. We hope that this summary will help the reader to rapidly navigate through the thesis, while understanding the relations between its different components.

Table of content

Synopsis
Part I: Introduction
Chapter 1: Brain Imaging
Chapter 2: Physiological Models of the hemodynamic response
Chapter 3: State of the Art of EEG/MEG-fMRI Fusion
Part II: Using Nonlinear Models in fMRI
Chapter 4: Model Identification
Chapter 5: Nonlinear Hypothesis Testing and Model Selection
Part III: Investigation on the Cerebral Blood Flow in Optical Imaging
Chapter 6: Dynamics and Nonlinearities of the Flow Response
Chapter 7: Bi-dimensional Flowmetry with Intrinsic Recordings
Part IV: Using Nonlinear Models in EEG-fMRI Fusion
Chapter 8: EEG-fMRI Fusion using Kalman Filtering
General Conclusion
Bibliography

Statistiques de consultation

Repository Staff Only: edit this item